Meta released Muse Wonder and Muse Wonder Full on October 12, 2026, and calls them its most capable models yet for personal and professional agent work.
Meta introduces Muse Wonder and Muse Wonder Full, frontier models for agents that handle personal and professional work. Meta calls them its most capable models to date. It says they reach state-of-the-art results on tasks such as an insurance claim, a lease, personal finances, and building a working app or game. The launch builds on Muse Spark, fourteen months after Meta began a ground-up overhaul of its AI efforts.
Two ideas define the series. Active Memory carries what matters about a person across conversations and brings back details relevant to the task. Active Focus keeps working when a task gets hard: the model tries other approaches, recovers from errors, asks for clarification, and checks in before anything important. Meta says both models do this in law, finance, and health. They were also trained on long software tasks so they can track requirements, change a codebase, and revise through testing.
Muse Wonder is available today for developers through the Meta Model API and Muse Code. Muse Wonder Full is available today for Muse Code subscribers. Muse Wonder will soon power experiences in Muse at muse.ai, first for subscribers, including artifact generation. Later, Meta plans to use it to build specialized models for everyone on Muse. Training ran at the gigawatt-scale Prometheus data center.
Meta says a personal agent may see messages, calendars, and documents, so the Muse Wonder family is trained not to share private information without permission and to refuse help with actually harmful acts. The company calls these its strongest safeguards yet against chemical, biological, radiological, nuclear, and cyber misuse, while developers can still look for flaws in their own code. The release follows the Meta Superintelligence Scaling Framework. A fuller safety report is due soon. Named tests include Harvey's Legal Agent Benchmark, FocusBench, and SWE-Marathon.
Read the announcement on the Meta Newsroom and the technical write-up on the Meta research blog.
Original Sources
Disclaimer
This is an AI-summarized article from authoritative sources. If you want your content removed, please .
